Effective optimization of emitters and surface passivation for nanostructured silicon solar cells

Abstract

Using a black silicon surface is a promising way to minimize the optical loss of solar cells; however, the strength of low optical loss is partially diminished due to an increase in surface recombination of nanostructured silicon surfaces. In this paper, we study the recombination mechanism of nanostructured silicon surfaces. Experimental results show that the loss in efficiency of nanostructured silicon solar cells is greatly dominated by the increased surface recombination and Auger recombination. In order to suppress the carrier recombination through the nanostructured surfaces, we developed a technique to modify the surface morphology and the doping concentration of the emitter. By adopting an optimized SiO₂/SiN_X passivation scheme, we obtained a compromise between low emitter recombination velocities and low reflectance. Remarkable gains of 3.7% on average efficiency, 34 mV on open circuit voltage, 3.65 mA cm⁻² on short circuit current density, and 4.65% on FF were obtained, comparing with the black silicon solar cells fabricated by a standard industrial process. A best solar cell of 18.5% efficiency was achieved.

---

1. Introduction

Reducing surface reflectance to enhance light absorption plays a significant role in improving the efficiency of solar cells. Most commonly, depositing anti-reflection coatings (ARCs) on micro-textured silicon is frequently applied to reduce reflectance.¹ However, ARCs based on quarter-wave-length only perform well at a specific wavelength and at specific angles of incidence.^2–4 Recently, nanostructured black silicon (b-Si) have attracted much attention because of their nearly ideal photo-absorption across the solar spectrum without ARCs. The nearly ideal photo-absorption is obtained by forming an effective medium with gradually varying refractive index, so that the incident light does not encounter abrupt index change from air to Si.^5–8 Many techniques, such as metal-assisted wet-chemical etching (MACE),^2,5,6,9–16 laser-induced etching,^17,18 reactive-ion etching^19–21 and electro-chemical etching²² have been proposed for fabricating nanostructured b-Si. The main challenges that hinder the application of nanostructured surface in solar cells are related to the increasing recombination due to the large surface area and high doping concentration.^13,23,24 Hence, the loss caused by minority-carrier recombination is one of the critical factors influencing the efficiency of black silicon solar cells. Generally, recombination in b-Si can be reduced by the following approaches: (1) decreasing the surface area; (2) reducing the interfacial state density, which can be achieved by decreasing the doping concentration near surfaces and using effective chemical passivations; (3) reducing the minority carrier concentration near the interface by built-in electric field, which is provided by fixed charges in the surface passivation layers.²⁵

Various materials, such as SiN_X, SiO₂, Al₂O₃ and their stacked forms are usually chosen as passivation layers to reduce the recombination of b-Si solar cells.^{13,23,26–28} Al₂O₃ films prepared by atomic layer deposition (ALD), which possess of negative fixed charges at the Al₂O₃/Si interface, can provide superior passivation quality for p⁺ emitters of n-type solar cells.^29–31 Savin et al. presented a 22.1% efficient black silicon solar cell based on n-type interdigitated back contact solar cells structure, the front surfaces of these cells was passivated by ALD Al₂O₃.²⁶ For phosphorus diffused n⁺ emitters of p-type b-Si solar cells, SiN_X films deposited by plasma enhanced chemical vapor deposition (PECVD) have become the most commonly industrial method used to provide the passivation and antireflection for silicon solar cells. Recently, our research group has developed a 19.0% efficient 241.2 mm²-large micro-nano dual-scale surface textured black silicon solar cells using SiN_X films as a passivation and antireflection layer.²³ However, the effect of the PECVD-deposited SiN_X layer is limited in high aspect ratio black silicon case because the SiN_X particles are difficult to be deposited into the deep and narrow holes of high aspect ratio black silicon.²⁷ Alternately, thermally grown SiO₂ has been widely applied to passivate n⁺ emitters of b-Si solar cells owing to its excellent chemical passivation for the Si–SiO₂ interface. Oh et al. used this approach to fabricate an 18.2% p-type black silicon solar cell in which the thermal SiO₂ passivation layer provided a very low effective surface recombination velocity of 53 cm s⁻¹ on b-Si surface; however, the efficiency of b-Si solar cells fabricated by Oh et al. is still lower than the cells with industry-standard pyramid textures and antireflection coatings, due to the insufficient field-effect passivation of SiO₂.¹³ Consequently, SiO₂/SiN_X stacked layer is suggested for the passivation of b-Si solar cells. This stack scheme can combine the good surface conformality of SiO₂ and the effective hydrogenation of dangling bond interface states during the SiN_X deposition, therefore provides higher passivation quality for b-Si solar cells.^28,32–35

In this paper, we focus on the optimization of the emitters and the passivation for p-type nanostructured b-Si solar cells. MACE process was adopted to fabricate nanostructures on silicon wafers. In order to reduce the recombination of nanostructured surfaces, we developed a surface morphology modification technique followed by an emitter optimization. The surface morphology modification was performed by a 0.5 wt% KOH etching after MACE, and the effective lifetime of the wafers etched with KOH solution for different time were analyzed to investigate the recombination mechanism of b-Si. To suppress the surface recombination velocity, a second KOH treatment was performed to optimize the emitters of b-Si after the phosphorus diffusion and SiO₂/SiN_X stacked passivation layer were used for the b-Si emitters. A significant efficiency improvement was achieved on a 156 × 156 mm² nanostructured b-Si solar cells.

2. Experimental

2.1 Preparation and optimization of b-Si wafers

P-Type Czochralski (CZ) mono-crystalline silicon wafers with a thickness of 200 ± 10 μm, and a resistivity of 1–3 Ω cm were used in this work. The wafer area was an industrial standard size of 156 × 156 mm². Saw damage was etched off in a 30% KOH for 1 min at 80 °C. Then wafers were cleaned in a mixed solution of 10 wt% HF and 10 wt% HCl at room temperature for 10 min and rinsed using deionized (DI) water. The silicon nanostructures were fabricated using MACE technique. In this process, silver nanoparticles were electrolessly deposited on the surface of wafers by immersing wafers in a mixed solution of 0.001 mol L⁻¹ AgNO₃ and 1 vol% HF solution for 90 s, then Si nanostructures were formed after being immersed in another mixed solution of 15 vol% HF, 3 vol% hydrogen peroxide (H₂O₂) and DI water for 25 s. Finally, the nanostructured surface silicon wafers were immersed into the silver etchant solution (NH₄OH : H₂O₂ : DI = 1 : 1 : 5, 10 min) to remove silver nanoparticles.

To modify the surface morphology, a 0.5 wt% KOH solution was used to change the formed nanostructures. The process flow for nanostructured surface morphology modification is described in Fig. 1(a). All samples are b-Si wafers which were subjected to KOH surface morphology modification after the MACE. This sample group was named as group A. The etching processes were carried out at room temperature for five different time conditions: 0 s, 15 s, 30 s, 45 s and 60 s. Afterwards, all wafers were cleaned in a mixed solution of HF and HCl and rinsed using DI water. Phosphorus diffusion using liquid POCl₃ was performed at a temperature of 850 °C to form the n-type emitters with a sheet resistances (R_sheet) of 80 Ω sq⁻¹. Then, the phosphosilicate glass (PSG) layer, formed during diffusion, was removed in a 5 wt% HF solution. At last, a SiO₂ passivation layer was formed by a dry thermal oxidized at 840 °C for 1800 s.

Fig. 1 Fabrication processes of the test samples used for the study of (a) b-Si surface morphology modification (group A), (b) b-Si emitter optimization (group B), (c) b-Si passivation schemes (group C). All test samples are of n⁺pn⁺ symmetrical structures.

We also fabricated samples with additional KOH modification for n⁺ emitters. The corresponding experimental process is shown in Fig. 1(b), and this sample group was named as group B. After the fabrication of b-Si surface, the surface morphology modification was performed in a 0.5 wt% KOH solution for 15 s. Afterwards, n⁺ emitters (R_sheet ≈ 50 Ω sq⁻¹) were formed in a liquid POCl₃ tube furnace at a temperature of 850 °C. The PSG layer was removed by a 5 wt% HF solution. To decrease the phosphorus concentration near the front surface, the phosphorus-diffused b-Si wafers were immersed in a 0.04 wt% KOH solution for 15 s, 30 s, and 60 s, respectively. The wafers were cleaned in a mixed solution of 5 wt% HF and 5 wt% HCl, and rinsed using DI water. Finally, all the cleaned wafers were oxidized at 840 °C for 1800 s to form a SiO₂ passivation layer on the surfaces.

The fabrication process for the samples used in the study of passivation is shown in Fig. 1(c). The SiN_X layers were deposited using PECVD method. In order to investigate the passivation quality of thermal SiO₂ as well as SiO₂/SiN_X dual-layer stacks as the passivation layer of optimized b-Si solar cells, n⁺pn⁺ symmetrical test structure with double-sided nano-structured surfaces were prepared. B-Si wafers with a KOH etching of 15 s after MACE and an additionally KOH modification of 30 s after diffusion were chosen as an optimized condition to investigate the passivation schemes of b-Si. After optimizing the emitters with KOH solution, the both sides of b-Si wafers were either coated with thermal SiO₂ or with SiO₂/SiN_X dual- layer stacks. This sample group, which is for the study of passivation schemes, was named as group C hereafter.

2.2 Fabrication of the b-Si solar cells

To fabricate the b-Si solar cells, MACE technique was applied to the as-cleaned silicon wafers to form b-Si nanostructures. The wafers were then divided into four groups. The wafers of group S1 followed the steps of standard process, which doesn't have any KOH etching in the preparation. After a standard cleaning process, the wafers were subjected to a phosphorus diffusion at 850 °C. The sheet resistance of the formed n⁺ emitter was 80 Ω sq⁻¹. The nanostructures on the rear side of wafers were subsequently etched off by a single-side wet etch in a mixed solution of HF and HNO₃ and followed by PSG removal. Then, a 10 nm thick thermal SiO₂ layers was formed in dry O₂ at 840 °C for 1800 s. Finally, the front and back metallization were carried out using industrial screen-printing and co-firing process. Wafers in group S2 followed the same process as that for group S1, except that a 45 s KOH etching was supplementally carried out after the MACE step to modify the surface morphology.

Wafers in group S3 were etched in a 0.5 wt% KOH solution at the room temperature for 15 s to modify the surface morphology after the MACE. Then, phosphorous diffusion was performed to form the n-type emitter with a sheet resistance of 50 Ω sq⁻¹. After PSG removal and back-side polished, the emitter optimization was carried out in a 0.04 wt% KOH solution for 30 s. After cleaning in HF solution, a 10 nm thick thermal SiO₂ layer was formed in dry O₂ at 840 °C for 1800 s. Finally, the front and back metallization were carried out using industrial screen-printing and co-firing process. Wafers in group S4 followed the same process as that for group S3, and additionally a 45 nm thick SiN_X layer was deposited on the front surface at a substrate temperature of 400 °C. A SiO₂/SiN_X dual-layer stacks was formed to passivate the front surface of b-Si.

2.3 Characterization

The morphologies of the b-Si were investigated using a Zeiss ΣIGMA-HD field emission scanning electron microscope (FESEM). The reflectance curves were measured with a Varian Cary 6000i spectrometer with an integrating sphere. The effective minority carrier lifetime was measured by quasi-steady-state photoconductance (QSSPC) instrument (WCT-120 by Sinton Consulting). The doping profiles of phosphorus were determined with a WEP CPV21 wafer profiler by electrochemical capacity voltage (ECV) profiling. The current–voltage (I–V) measurements of solar cells were carried out using Berger PSS 10-S test system.

3. Results and discussion

The SEM images in Fig. 2 compare the surface morphology of wafers in group A. Fig. 2(a) shows the SEM image of nanostructures of b-Si without KOH etching. We see that the nanopores are vertically aligned cylinders with the heights ranging from 400 to 500 nm, and the diameters are ranging from 50 to 100 nm. The SEM images of nanostructured surface treated in 0.5 wt% KOH solution for 15 s, 30 s, 45 s, and 60 s are shown in Fig. 2(b)–(e), respectively. It is clearly seen that the average diameter of the nanopores increases with the treating time, while the height of the nanopores inversely proportional to the treating time. As increasing the KOH etching time from 0 s to 15 s and 30 s, the vertically aligned cylinders morphology of nanostructured surface become inverted conical-like hillocks, as shown in Fig. 2(b) and (c). Further increase of KOH etching time to 45 s and 60 s leads to a spectacular change on the surface: instead of conical-like hillocks, inverted pyramids are formed.

Fig. 2 SEM images of nanostructured silicon with different KOH etching time after MACE: (a) 0 s, (b) 15 s, (c) 30 s, (d) 45 s, (e) 60 s. The inset image is a top view of the nanostructured silicon. (f) Calculated enlargement area factor A_Nano/A_Flat and the solar-spectrum-weighted average reflectance of b-Si as a function of KOH etching time after MACE.

In order to investigate the complex recombination mechanism from the lifetime measurement, we introduced a ratio factor to describe the enlargement in nanostructured surface area. The enlargement area factor can be defined as A_Nano/A_Flat, where A_Nano is the effective area of nanostructured surface and A_Flat is the area of flat surface. The values of A_Nano/A_Flat for the b-Si wafers in Fig. 2(a)–(e) are calculated to be 8.5, 4.2, 3.3, 2.6, and 2.1. We can see from Fig. 2(f) that the A_Nano/A_Flat of nanostructured surface sharply decease from 8.5 to 4.2 after 15 s KOH etching, and then slowly decrease to 2.1 after an additional 45 s etching. The right axis of Fig. 2(f) describes the variation of average reflectance (R_ave) versus KOH etching time. R_ave is the solar-spectrum-weighted average reflectance of b-Si measured over the entire wavelength range between 300 to 1100 nm. The R_ave of the b-Si without KOH etching is calculated to be 3.41%, and increases to 12.87% after 60 s KOH etching. It is known that the nanopores on b-Si surface with random depths and diameters can be considered as a set of effective mediums with gradually varying refractive index between air (1) and Si (3.5), which suppresses the reflectance effectively.^36,37 As shown in Fig. 2(a)–(e), as the KOH etching time increases, the nanopores become larger and shallower, which means that the number of medium layers between air and Si decreases, and thus results in the increase of R_ave of b-Si.

Fig. 3(a) plots the minority carrier lifetime as a function of injection level for b-Si samples in group A, which were etched by KOH for 0 s, 15 s, 30 s, 45 s, and 60 s after MACE. The test wafers are symmetrical structure with nano-structures, phosphorus diffusion emitter and thermal SiO₂ passivation on both sides. At an injection level of 1 × 10¹⁵ cm⁻³, the effective lifetimes are measured to be 6 μs, 33 μs, 47 μs, 63 μs, and 72 μs for b-Si with KOH treatment of 0 s, 15 s, 30 s, 45 s, and 60 s, respectively. The curves in Fig. 3(a) show that the minority-carrier lifetime increases with KOH etching time.

Fig. 3 (a) Minority carrier lifetimes as a function of injection level for b-Si with 0 s, 15 s, 30 s, 45 s, and 60 s KOH etching after MACE. (b) S_eff as a function of A_Nano/A_Flat (bottom x-axis) and KOH etching time (top x-axis). S_loc for all samples is also present in figure. The insert graph is a zoom of the first four points and the dotted line are guides to eyes showing the increase trend.

Effective surface recombination velocity (S_eff) is one of the most important parameters of solar cells which directly reflects the minority carrier recombination behavior at the surface. In order to further investigate the recombination mechanism of b-Si, we calculated the S_eff for the five test samples in group A. At low-level injection condition, S_eff can be calculated from the measured effective lifetime, using the approximation given by Sproul:³⁸

where τ_bulk is the bulk Shockley–Read–Hall (SRH) lifetime, S_eff(front) and S_eff(back) are the effective surface recombination velocities at the front and back surface respectively, and W is the wafer thickness. For our symmetrical test structures, both sides of wafer are assumed to be etched and passivated identically, i.e. S_eff = S_eff(front) = S_eff(back), one obtains:³⁹

We use 300 μs for the τ_bulk value for the p-type Cz wafers in our experiments, and assume that it does not change with the injection conditions. S_loc is defined as the per unit area effective surface recombination velocity, which includes the recombination at the surface and the doped region near the surface. S_loc is introduced in order to separate the complex recombination courses of the nanostructured surface. It can be calculated by the model of Oh et al.:¹³

Fig. 3(b) shows the S_eff as a function of A_Nano/A_Flat and KOH etching times. S_loc calculated by eqn (3) is also plotted as a function of A_Nano/A_Flat and KOH etching time, as shown in Fig. 3(b). If it is presumed that the increase of surface recombination velocity is only related to the enlarged surface area, the S_loc values should be approximately identical for all our nanostructured samples. However, the results extracted from experiments are not in this case. It is found that, the S_loc value decreases from 192 cm s⁻¹ to 62 cm s⁻¹ after 15 s KOH etching, and decreases to 54 cm s⁻¹ after an additional 15 s KOH etching, as shown in Fig. 3(b). This result indicates that, for high A_Nano/A_Flat, the decrease of S_eff is not only related to the reduction of surface area but also related to the reduction of the recombination near the surface. The S_loc values do not change significantly after 45 s, and 60 s KOH etching, which reveals that surface recombination dominates the mechanism for b-Si with low A_Nano/A_Flat.

For high aspect ratio black silicon wafers, more surface area was exposed to the atmosphere in furnace during the diffusion process. Assuming that the phosphorus concentration per unit surface area are identical for all the wafer samples during the diffusion process, more phosphorus atoms can reach and diffuse through the nanostructured surface for nanostructured samples of high A_Nano/A_Flat compared with the micro-textured surface samples. On our experiment condition, the junction depth after diffusion is more than 400 nm. This value generally exceeds the 50–200 nm thick walls between nanopores. Consequently, the doping concentration of phosphorus near the surfaces of high A_Nano/A_Flat wafers would be much higher than those with low A_Nano/A_Flat. Therefore, strong Auger recombination in such emitters of high phosphorus surface concentration (higher than 1 × 10²⁰ cm⁻³) is a quite dominant reason leads to the decrease of τ_eff for b-Si samples.⁴⁰

As discussed above, the effective lifetime τ_eff of b-Si can be enhanced by KOH etching after MACE attributed to the reduction of surface area. It is known that the region with very high concentration of phosphorus near the surface, which is usually called as dead layer,⁴¹ can cause a large S_eff and poor electric performances of solar cells. The region of dead layer can be minimized by optimizing the parameters in diffusion process or by performing a wet etching after the diffusion. For b-Si samples, performing KOH modification after the diffusion is a controllable etching method to shorten the dead layer of b-Si. However, an over decrease in dead layer depth possibly leads to an increase in the specific contact resistance. For instance, if we suppress the S_eff below 75 cm s⁻¹ with b-Si samples without morphology treatment before diffusion, the sheet resistance need be etched to over 120 Ω sq⁻¹ due to the large A_Nano/A_Flat value. The sheet resistance will be too high to form a good ohmic contact between silver pastes and the front emitters. Considering the two aspects, the following experiments were carried out on the b-Si wafers which are previously etched with KOH after MACE.

In order to further study the effect of a second time KOH etching after diffusion process, we chose the b-Si samples etched with KOH after MACE for 15 s, and performed a second etching in 0.04 wt% KOH solution for 15 s, 30 s, and 60 s after POCl₃ diffusion, as described in Experimental section for group B. Fig. 4(a)–(c) present the cross-sectional SEM images of the samples in group B. It shows that the nanopores of the 15 s KOH etching b-Si are 300–350 nm in depth and 100–200 nm in diameter. As can be seen in Fig. 4(b), when the KOH modification time increases to 30 s, the inverted conical-like hillocks on nanostructured surface become inverted pyramids. The depth and the diameter are measured to be 250–300 nm and 150–250 nm, respectively. As shown in Fig. 4(c), further increase of KOH modification time to 60 s, nano-scale random pyramids with a height of 200–400 nm appear, due to the anisotropic etching property of KOH solution.

Fig. 4 SEM images of nanostructured silicon with 15 s KOH etching after MACE and a second etching with 0.04% KOH for different time after phosphorus diffusion: (a) 15 s, (b) 30 s, (c) 60 s. The inset image is a top view of the nanostructured silicon.

Fig. 5 presents the doping profiles of the b-Si samples in group B. For the sample with 15 s KOH solution modification in group B, the peak doping concentration (1.1 × 10²¹ cm⁻³) is lower and the junction depth (∼0.49 μm) is slightly shorter than those of the samples without modification after diffusion (surface peak doping concentration ∼2.5 × 10²¹ cm⁻³, junction depth ∼0.5 μm). The peak doping concentration decreased to 2.8 × 10²⁰ cm⁻³ and 1.8 × 10²⁰ cm⁻³ after 30 s and 60 s KOH modification, respectively. It is clear that increasing the KOH modification time leads to a decrease in peak doping concentration and junction depth. As a result, the sheet resistance increases from 50 Ω sq⁻¹ to 65 Ω sq⁻¹, 80 Ω sq⁻¹, and 100 Ω sq⁻¹ after 15 s, 30 s and 60 s KOH etching, respectively. The decrease of surface phosphorus concentration by KOH etching after the diffusion signifies that the Auger recombination in b-Si emitter declines. The doping profile of the b-Si sample with a KOH etching for 45 s after MACE is also plotted in Fig. 5 for comparison. Although the measured sheet resistance of the sample in group A is similar to the sample with 30 s KOH etching in group B, we can see that their doping profile were obviously different. The phosphorus concentration at the surface-near emitter is higher for b-Si sample in group A. Because the diffusion depth generally exceeds the thickness of walls between nanopores, the phosphorus doping concentration for b-Si wafers would reach very high concentration near the surface. Therefore, the use of a second KOH modification after diffusion is necessary to reduce the depth of dead layer.

Fig. 5 Phosphorus doping profiles for b-Si wafers treated with KOH. The sample in group A is a b-Si wafer with a 45 s KOH etching. The sheet resistance is 80 Ω sq⁻¹. The samples in group B are b-Si wafers with a 15 s KOH etching, and then suffered a second KOH modification for 0 s, 15 s, 30 s, and 45 s after diffusion. Their corresponding sheet resistances are 50, 65, 80 and 100 Ω sq⁻¹ measured by four-point probe method.

The passivation effects of SiO₂ on KOH modified emitters are then studied by lifetime measurements. The injection level dependent minority carrier effective lifetime for the b-Si in group B is given in Fig. 6(a). At an injection level of 1 × 10¹⁵ cm⁻³, the lifetime is 31 μs for b-Si without etching after diffusion, and it increases to 56 μs, 76 μs, and 98 μs for b-Si with 15 s, 30 s, and 60 s KOH modification respectively. The passivation properties of SiO₂ are compared between the samples with only KOH etching after MACE and the samples with KOH etching after MACE plus a second KOH modification after diffusion. Fig. 6(b) shows the S_eff as a function of A_Nano/A_Flat for b-Si wafers in group B. For comparison, the S_eff values versus A_Nano/A_Flat of samples in group A are also plotted. It is clearly seen that the S_eff of wafers both in group A and group B increases proportionally with the surface enlargement area factor A_Nano/A_Flat. Moreover, the S_eff values of b-Si in group B are larger than those in group A when A_Nano/A_Flat is larger than 3.2, because of lower sheet resistance and thus higher Auger recombination velocity. However, the S_eff values of b-Si in group B are smaller than those in group A when the A_Nano/A_Flat is smaller than 3.2. The reason is that one single KOH etching after MACE is not sufficient to decrease the doping concentration near the nanostructured surface, a second KOH modification process after diffusion was used to provide a better modification of the doping concentration near the surface. The surface recombination velocity is consequently reduced.

Fig. 6 (a) Minority carrier lifetimes as a function of injection level for samples in group B. The wafers are b-Si without KOH modification, as well as, b-Si with 15 s, 30 s, and 45 s KOH modification after diffusion. (b) Comparison of S_eff as a function of A_Nano/A_Flat for samples in group A and group B.

For comparison, the standard b-Si solar cells without any etching (group S1), the b-Si solar cells with a 45 s KOH etching after MACE (group S2) and the b-Si solar cells with a first 15 s KOH etching and a second KOH modification for 30 s after diffusion (group S3) were fabricated. The sheet resistance of the emitters of all cells were 80 Ω sq⁻¹, and all cells were passivated by SiO₂. The J–V characteristics were summarized in Table 1. Compared to the b-Si solar cells in group S1, all electrical parameters of the b-Si solar cells in group S2 were significantly improved. The V_OC of solar cells in group S2 showed an improvement by 10 mV compared to the solar cells in group S1, attributing to the suppression of the surface recombination and the emitter Auger recombination. The average FF of the group S2 presented a satisfactory value of 77.17% which is higher by absolute 3.96% than that of the b-Si solar cells in group S1. The relatively lower FF for the group S1 might be resulted from poor ohmic contact between the silver pastes and the nanostructured surface, since it is difficult for the silver pastes to reach the bottom of nanopores on unmodified surfaces using screen-printing method.⁴² While this problem has been significantly reduced using surface morphology modification, in which the nanostructure morphology is smoother for the screen-printing process to form a good ohmic contact between the silver contacts and Si nanopores. Finally, the b-Si solar cells in group S2 exhibited an improvement in J_SC of 2.87 mA cm⁻² compared to the b-Si solar cells in group S1 (33.24 mA cm⁻²), due to the reduction of the carrier recombination loss and the improved FF. As a result, the absolute η of b-Si solar cells in group S2 is improved 2.4% on average compared to the b-Si solar cells in group S1.

Table 1 The electrical characteristics of the standard b-Si solar cells without any etching (group S1), the b-Si solar cells with a 45 s KOH etching after MACE (group S2) and the b-Si solar cells with an additional 30 s KOH modification after diffusion (group S3). The average values are calculated from 4 to 12 cells

Cells		VOC (mV)	JSC (mA cm^−2)	FF (%)	η (%)
S1	Average	605 ± 4.0	33.24 ± 0.69	73.21 ± 1.38	14.7 ± 0.4
	Best	608	34.01	74.16	15.3
S2	Average	615 ± 4.5	36.11 ± 0.20	77.17 ± 0.57	17.1 ± 0.2
	Best	617	36.32	77.28	17.3
S3	Average	623 ± 5.8	36.30 ± 0.35	77.24 ± 0.38	17.5 ± 0.3
	Best	626	36.47	77.53	17.7

---

The electrical performances of solar cells can be further improved using an additional KOH modification after diffusion. As shown in Table 1, the conversion efficiency of the b-Si solar cells in group S3 was improved by 0.4% compared to the b-Si solar cells in group S2. The V_OC was improved by 8 mV and the J_SC of the solar cells in group S3 was also slightly higher than that in group S2. It can be seen that the second KOH modification performed an important role in the suppression of the surface recombination and the Auger recombination in the emitter. Although the cells in group S3 were not of the lowest surface reflectance among the three groups, they still gained the best conversion because of lower carrier recombination loss.

The passivation quality can be further developed using SiO₂/SiN_X stacks. Fig. 7 displays the SEM image of SiO₂/SiN_X stacks deposited on the KOH modified emitters. The thickness of the thermal SiO₂ layer is about 10 nm. The SiN_X films, deposited on top of the thin SiO₂ is about 45 nm. From the SEM image we see that the nanostructured surface of b-Si is very well wrapped by the SiO₂/SiN_X stack layer, which is attributed to the modification of surface structure and our previously optimized SiN_X film deposition technique. The thickness of SiN_X film is optimized to be 45 nm, to compromise the high quality passivation and the parasitic loss between 300 and 400 nm in absorption spectrum.⁴³

Fig. 7 SEM image of a nanostructured silicon surface. The sample was etched by KOH for 15 s after MACE and a second KOH modification for 30 s after diffusion. The passivation scheme is SiO₂/SiN_X stacks. The white thin film between Si substrate and the SiN_X layer is SiO₂. The thickness of the thermal SiO₂ film is about 10 nm, the thickness of SiN_X layer, deposited on top of the thin SiO₂ is about 45 nm.

Fig. 8 shows injection level dependent effective lifetime curves for b-Si passivated by thermal SiO₂ films and the dielectric stack of SiO₂/SiN_X films. The results show that the optimized b-Si wafers with SiO₂/SiN_X passivation have a superior effective lifetime of 140 μs, compared to the optimized b-Si wafers only with SiO₂ passivation have an effective lifetime of 76 μs. In Fig. 8, the S_eff values calculated using eqn (2) for the two types of samples are also shown.

Fig. 8 Minority carrier lifetime as a function of injection level. The b-Si samples are from group C, which are etched by KOH for 15 s after MACE and then with KOH modification for 30 s after diffusion. The surfaces are passivated with SiO₂ and SiO₂/SiN_X stacks. The S_eff values are also shown.

By performing a SiO₂/SiN_X stack passivation for optimized b-Si wafers, a low S_eff of 38 cm s⁻¹ was obtained, which is a remarkable improvement compared to the 98 cm s⁻¹ of those using only SiO₂ passivation. This improved passivation properties for b-Si can be attributed to the high quality of the Si/SiO₂ interface, and the subsequent transport of hydrogen atoms from the SiN_X film to the Si/SiO₂ interface, which leads to the saturation of dangling bonds at the Si/SiO₂ interface.^32,44–46

From the experimental results and the discussion above, we applied the emitter modification process and optimized passivation to our b-Si solar cells. Fig. 9 shows the schematic cross-section of the b-Si solar cells. At the front side, the nanostructured surface is applied to reduce the reflection of silicon surface. Surface and emitter modification by 15 s KOH etching after MACE and additionally 30 s KOH modification after the diffusion were carried out to eliminate the surface and Auger recombination induced by excessive doping through the high surface area of b-Si. The emitter was passivated by the dielectric stack of SiO₂/SiN_X films. The front and back metallization of wafers were carried out using commercial screen-printing and co-firing process.

Fig. 9 Sketch of the b-Si solar cell passivated by SiO₂/SiN_X stacks.

Table 2 shows the J–V characteristics of the b-Si solar cells with two-time KOH optimization and SiO₂/SiN_X stacked passivation. The electrical performances of solar cells are further improved using SiO₂/SiN_X stacks. Comparing the experimental results of the optimized b-Si solar cells passivated by SiO₂/SiN_X stacks with those passivated by SiO₂, we found that the V_OC, J_SC, FF, and η of the stacks passivation are further enhanced by 16 mV, 0.59 mA cm⁻², 0.62% and 0.9% respectively. A higher value for V_OC is achieved due to the great suppression of recombination imposed by the SiO₂/SiN_X stacks passivation. The J_SC of stacks passivation solar cells is higher with an absolute increment of 0.59 mA cm⁻², owing to lower reflectance and the suppression of surface recombination after SiN_X deposition. The slightly increased fill factor may be explained by that Ag pastes have a better match to the SiN_X surface. The b-Si solar cells passivated by SiO₂/SiN_X present 0.9% absolute increase in average efficiency compared with those passivated by SiO₂. The I–V characteristic of the best nanostructured b-Si solar cells is shown in Fig. 10. A highest efficiency of 18.5% (V_OC = 640 mV, J_SC = 37.02 mA cm⁻², FF = 78.04%) was achieved.

Table 2 The electrical characteristics of the optimized b-Si solar cells passivated SiO₂/SiN_X stacks (group S4). The average values are calculated from 4 to 12 cells

Cells		VOC (mV)	JSC (mA cm^−2)	FF (%)	η (%)
S4	Average	639 ± 0.9	36.89 ± 0.07	77.86 ± 0.35	18.4 ± 0.1
	Best	640	37.02	78.04	18.5

---

Fig. 10 I–V Characteristic of the best-optimized 156 × 156 mm² crystalline solar cells.

4. Conclusions

In this paper, the recombination mechanism and the surface modification technique to suppress the recombination of b-Si are investigated. It is shown that the decrease in effective lifetime is not only due to the enlarged surface area but also due to the Auger recombination caused by the excessive doping through the high surface area of b-Si. We found that the effective lifetime improves with KOH etching time after MACE and after diffusion, due to the decrease of surface area and the suppression of Auger recombination. B-Si wafers with a 15 s KOH etching after MACE and additional 30 s KOH modification after the diffusion exhibit better performance as a compromise between low surface recombination and low reflectance. We presented an effective passivation scheme for the nanopore structures of b-Si surfaces fabricated via MACE. By performing the SiO₂/SiN_X passivation for our optimization b-Si, a very low effective surface recombination velocity of 38 cm s⁻¹ has been achieved. Incorporated the stack passivation scheme, the surface morphology modification and emitter optimization technique, the black silicon solar cells obtained a large improvement of 3.7% in average η compared to standard black silicon solar cells without any optimization. Finally, a highest η of 18.5% with a V_OC of 640 mV, J_SC of 37.02 mA cm⁻², and FF of 78.04% were achieved on a 156 × 156 mm²-large p-type black silicon solar cell.

Acknowledgements

The authors would like to thank Solargiga Energy Holdings Ltd. for providing p-type Cz silicon wafers and the PECVD passivation support. This work has been partially funded by the Doctoral Scientific Research Foundation of Liaoning Province No. 201501180, the Fundamental Research Funds for the Central Universities under Contracts No. DUT16LK28 and SRF for ROCS, SEM.

References

1. P. Doshi, G. E. Jellison and A. Rohatgi, Appl. Opt., 1997, 36, 7826–7837.
2. S. Koynov, M. S. Brandt and M. Stutzmann, Appl. Phys. Lett., 2006, 88, 203107.
3. H. Nagel, A. G. Aberle and R. Hezel, Prog. Photovoltaics, 1999, 7, 245–260.
4. A. Premoli and M. L. Rastello, Appl. Opt., 1994, 33, 2018.
5. H. M. Branz, V. E. Yost, S. Ward, K. M. Jones, B. To and P. Stradins, Appl. Phys. Lett., 2009, 94, 231121.
6. H.-C. Yuan, V. E. Yost, M. R. Page, P. Stradins, D. L. Meier and H. M. Branz, Appl. Phys. Lett., 2009, 95, 123501.
7. P. B. Clapham and M. C. Hutley, Nature, 1973, 244, 281–282.
8. R. B. Stephens and G. D. Cody, Thin Solid Films, 1977, 45, 19–29.
9. X. Li and P. W. Bohn, Appl. Phys. Lett., 2000, 77, 2572.
10. L. Tsakalakos, J. Balch, J. Fronheiser, B. A. Korevaar, O. Sulima and J. Rand, Appl. Phys. Lett., 2007, 91, 233117–2331173.
11. K. Nishioka, T. Sueto and N. Saito, Appl. Surf. Sci., 2009, 255, 9504–9507.
12. K. Hagedorn, C. Forgacs, S. Collins and S. Maldonado, J. Phys. Chem. C, 2010, 114, 12010–12017.
13. J. Oh, H.-C. Yuan and H. M. Branz, Nat. Nanotechnol., 2012, 7(11), 743–748.
14. Y. Wang, Y. P. Liu, T. Lai, H. L. Liang, Z. L. Li, Z. X. Mei, F. M. Zhang, A. Kuznetsov and X. L. Du, RSC Adv., 2013, 3, 15483.
15. L. Yang, Y. Liu, Y. Wang, X. Li, W. Chen, Y. Hua, Q. Zhang, J. Fu, H. Liang, Z. Mei and X. Du, RSC Adv., 2014, 4, 24458.
16. Y. T. Lu and A. R. Barron, ACS Appl. Mater. Interfaces, 2015, 7, 10288–10237.
17. X. Wang, F. Chen, H. Liu, W. Liang, Q. Yang, J. Si and X. Hou, Opt. Commun., 2011, 284, 317–321.
18. G. Nava, R. Osellame, R. Ramponi and K. C. Vishnubhatla, Opt. Mater. Express, 2013, 3, 612.
19. J. Yoo, Sol. Energy, 2010, 84, 730–734.
20. H. Jansen, M. De Boer, R. Legtenberg and M. Elwenspoek, J. Micromech. Microeng., 1995, 5, 115–120.
21. J. I. Gittleman, E. K. Sichel, H. W. Lehmann and R. Widmer, Appl. Phys. Lett., 1979, 35, 742–744.
22. C. C. Striemer and P. M. Fauchet, Appl. Phys. Lett., 2002, 81, 2980–2982.
23. P. Li, Y. Wei, Z. Zhao, X. Tan, J. Bian, Y. Wang, C. Lu and A. Liu, Appl. Surf. Sci., 2015, 357, 1830–1835.
24. C.-H. Hsu, J.-R. Wu, Y.-T. Lu, D. J. Flood, A. R. Barron and L.-C. Chen, Mater. Sci. Semicond. Process., 2014, 25, 2–17.
25. B. Hoex, J. J. H. Gielis, M. C. M. Van De Sanden and W. M. M. Kessels, J. Appl. Phys., 2008, 104, 1–7.
26. H. Savin, P. Repo, G. von Gastrow, P. Ortega, E. Calle, M. Garín and R. Alcubilla, Nat. Nanotechnol., 2015, 10, 1–6.
27. Y. Liu, T. Lai, H. Li, Y. Wang, Z. Mei, H. Liang, Z. Li, F. Zhang, W. Wang, A. Y. Kuznetsov and X. Du, Small, 2012, 8, 1392–1397.
28. Z. Zhao, B. Zhang, P. Li, W. Guo and A. Liu, Int. J. Photoenergy, 2014, 16, 7618–7626.
29. B. Hoex, S. B. S. Heil, E. Langereis, M. C. M. Van De Banden and W. M. M. Kessels, Appl. Phys. Lett., 2006, 89, 3–6.
30. B. Hoex, J. Schmidt, P. Pohl, M. C. M. van de Sanden and W. M. M. Kessels, J. Appl. Phys., 2008, 104, 044903.
31. K.-S. Jeong, S.-K. Oh, H.-S. Shin, H.-J. Yun, S.-H. Kim, H.-R. Lee, K.-M. Han, H.-Y. Park, H.-D. Lee and G.-W. Lee, Jpn. J. Appl. Phys., 2014, 53(4s), 853–871.
32. G. Dingemans, M. M. Mandoc, S. Bordihn, M. C. M. van de Sanden and W. M. M. Kessels, Appl. Phys. Lett., 2011, 98, 222102.
33. X. Liu, P. R. Coxon, M. Peters, B. Hoex, J. M. Cole and D. J. Fray, Energy Environ. Sci., 2014, 7, 3223–3263.
34. R. S. Bonilla, C. Reichel, M. Hermle and P. R. Wilshaw, J. Appl. Phys., 2014, 115, 144105.
35. Y. Larionova, V. Mertens, N.-P. Harder and R. Brendel, Appl. Phys. Lett., 2010, 96, 032105.
36. L. L. Ma, Y. C. Zhou, N. Jiang, X. Lu, J. Shao, W. Lu, J. Ge, X. M. Ding and X. Y. Hou, Appl. Phys. Lett., 2006, 88, 171907.
37. H. Lv, H. Shen, Y. Jiang, C. Gao, H. Zhao and J. Yuan, Appl. Surf. Sci., 2012, 258, 5451–5454.
38. A. B. Sproul, J. Appl. Phys., 1994, 76, 2851–2854.
39. S. Duttagupta, Ph.D. thesis, Department of Electrical and Computer Engineering, National University of Singapore, 2014.
40. B. Min, H. Wagner, A. Dastgheib-Shirazi and P. P. Altermatt, Energy Procedia, 2014, 55, 115–120.
41. B. Bazer-Bachi, E. Fourmond, P. Papet, L. Bounaas, O. Nichiporuk, N. Le Quang and M. Lemiti, Sol. Energy Mater. Sol. Cells, 2012, 105, 137–141.
42. Y. Tang, C. Zhou, S. Zhou, Y. Zhao, W. Wang, J. Fei and H. Cao, Chin. J. Chem. Phys., 2013, 26, 102.
43. A. Ingenito, O. Isabella and M. Zeman, Prog. Photovoltaics, 2015, 23, 1649–1659.
44. M. J. Kerr, Ph. D. thesis, The Austalian National University, 2002.
45. J. Schmidt, M. Kerr and A. Cuevas, Semicond. Sci. Technol., 2001, 16, 164.
46. Y. Wei, P. Li, Y. Wang, X. Tan, C. Song, C. Lu, Z. Zhao and A. Liu, Sol. Energy, 2015, 118, 384–389.

---